SPITFIRE-2: an improved fire module for Dynamic Global Vegetation Models

Abstract

Abstract. Fire is the primary disturbance factor in many terrestrial ecosystems. Wildfire alters vegetation structure and composition, affects carbon storage and biogeochemical cycling, and results in the release of climatically relevant trace gases, including CO2, CO, CH4, NOx, and aerosols. Assessing the impacts of global wildfire on centennial to multi-millennial timescales requires the linkage of process-based fire modeling with vegetation modeling using Dynamic Global Vegetation Models (DGVMs). Here we present a new fire module, SPITFIRE-2, and an update to the LPJ-DGVM that includes major improvements to the way in which fire occurrence, behavior, and the effect of fire on vegetation is simulated. The new fire module includes explicit calculation of natural ignitions, the representation of multi-day burning and coalescence of fires and the calculation of rates of spread in different vegetation types, as well as a simple scheme to model crown fires. We describe a new representation of anthropogenic biomass burning under preindustrial conditions that distinguishes the way in which the relationship between humans and fire are different between hunter-gatherers, obligate pastoralists, and farmers. Where and when available, we evaluate our model simulations against remote-sensing based estimates of burned area. While wildfire in much of the modern world is largely influenced by anthropogenic suppression and ignitions, in those parts of the world where natural fire is still the dominant process, e.g. in remote areas of the boreal forest, our results demonstrate a significant improvement in simulated burned area over previous models. With its unique properties of being able to simulate preindustrial fire, the new module we present here is particularly well suited for the investigation of climate-human-fire relationships on multi-millennial timescales.